Liquid Instruments, a developer of test and measurement solutions including software-reconfigurable hardware platforms, has integrated the ChatGPT language model into its Moku family of products. Moku devices, which are field programmable gate array (FPGA)-enabled, aim to emulate and eliminate bulky hardware such as oscilloscopes and spectrum analyzers, as well as advanced tools like a laser lockbox and lock-in amplifier.
To expedite FPGA programming, ChatGPT integrates with Moku Cloud Compile to create a feature that Liquid Instruments said allows users to run custom logic via a Moku device’s FPGA without any boilerplate or software download. Moku devices are programmed using VHDL, a language used to program functions for FPGA devices.
According to Liquid Instruments CEO Daniel Shaddock, the interoperability between Moku devices and ChatGPT supports users in a range of applications, from quantum optics to electronics research.
VHDL code, Shaddock said, can be time-intensive to write and presents a steep learning curve to users without prior knowledge of the language. The integration, he said, lowers the barrier to entry for VHDL coding. It also speeds testing and allows for faster deployment of custom code to real-world hardware meaning that users can accelerate development timelines and optimize workflows to save time and resources, Shaddock told Photonics Media.
The feature is especially useful for advanced experiments or product design initiatives with unique requirements. “Unlike traditional equipment, user-programmable FPGAs provide higher-performance solutions that can add custom functionality to an instrument and interact with real-world signals in real time,” Shaddock said.
According to Shaddock, Cloud Compile has built-in mechanisms that allow users to implement custom features without corrupting the device or causing it to fail. Shaddock posed an example in which an engineer is looking to lock their q-switched laser with a seed source to an optical cavity: They want the Moku device to generate the control signal from the laser lockbox during a specified locking window based on input and feedback from the source, and to hold the output at the previous value outside of the locking window, disregarding the input. The locking window can be indicated by a trigger signal.
“To enable the hold functionality for the laser lockbox, users can ask ChatGPT to create the required VHDL code, which can then be copied and pasted into Moku Cloud Compile,” Shaddock told Photonics Media. “Users can also apply this gating function to other applications such as spurious signal elimination, pulsed signal on-time measurement, and time-gated single-photon detection.”
Liquid Instruments plans to engage its customers to gather feedback on their use cases, and, Shaddock said, the company’s R&D team will continue to look for new ways to use ChatGPT to accelerate FPGA programming for a range of applications. These range from debugging code and anomaly detection to time-gated measurements and signal conditioning, Shaddock said.